Brain-machine interfaces (BMIs) promise to restore voluntary movement to paralyzed individuals by using their own neural activity to control movement of computer cursors, robotic arms, or even their own limbs. A class of biomimetic BMI research aspires to allow patients to control the BMI by imagining natural limb movements. However, as experimental monkeys transition from normal movements to controlling a BMI, the control neurons alter their discharge patterns in a manner indicative of motor learning. These changes, which occur over several sessions as the monkeys achieve peak performance, suggest there is nonetheless, a need to learn a new mapping between neural activity and the desired movements. The goal of my proposal is to improve BMI control through a better understanding of how and why the activity of cortical neurons changes during motor learning during both normal motor tasks and a BMI paradigm. I will investigate how two functionally distinct motor cortical areas, primary motor cortex (M1) and dorsal premotor cortex (PMd), work together to coordinate movements. A better understanding of the role of these areas in normal movements will allow us to design more biomimetic BMIs. Due to the apparent similarity between adapting to BMI use and traditional motor learning, I propose first to study how movement-related activity of motor cortical neurons changes during two common motor learning tasks. M1 and PMd are known to have different roles in reaching, with M1 activity that is closely related to limb dynamics and muscle activation and PMd encoding movement goals and planning. The different functional behavior of neurons in these areas suggests that they may differ in their utility for BMI control. I first propose to study how M1 and PMd alter their dischage as monkeys make reaching movements in the presence of two perturbations: a velocity-dependent force field (curl field) and a static visuomotor rotation. I will study how the two neural populations change to mediate adaptation using neural spatial tuning analysis methods as well as by inferring changes in functional connectivity between neurons. In the second aim, I will extend my experiments to include a BMI task to better understand the cortical changes observed during BMI use. Monkeys will use either M1 or PMd neurons to control a cursor over multiple sessions. As the monkeys adapt, I will characterize the changes in spatial tuning or functional connectivity underlying BMI skill acquisition. Additionally, I will test for what differences, if any, exist between M1 control and PMd control. The results of these two aims will help to understand the neural basis of motor learning, how M1 and PMd work together to coordinate movements, and how M1 and PMd might be used to create a more biomimetic BMI.